JPH095154A - Adaptive type directional sound detector - Google Patents

Abstract

PURPOSE: To provide a handy adaptive type directional sound detector with an improved S/N ratio by catching a sound from a noise source with a directional microphone. CONSTITUTION: A first directional acoustic sensor 1 is arranged as input signal microphone facing an apparatus to be detected with directivity. Then, a second directional acoustic sensor 2 which detects noises produced from wide-ranging noise sources without detecting the sound of the apparatus to be detected is arranged on the reverse side of the first directional acoustic sensor. After amplified by an amplifier 5, signals (an input signal 3 and a reference signal 4) from microphones are inputted into an A/D converter 6 to be digitized. The resulting signals are sent to an electronic computer 7 and noise components in the input signal are erased by an adaptive type noise canceler processing using an LMS algorithm in the computer 7 to be outputted as output signal 8.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for detecting a sound generated from a target apparatus with good S / N by reducing noise from the surroundings, and a noise canceling technology particularly suitable for acoustic equipment abnormality diagnosis The present invention relates to an adaptive directional sound detecting device using a.

[0002]

2. Description of the Related Art In an acoustic equipment abnormality diagnosis apparatus,
The noise from the surroundings is reduced, and the sound of the target device is S / N.
In order to detect well, a directional microphone such as a gun microphone type, a parabolic antenna type, or an array microphone type is widely used as a sensor that effectively detects the sound coming from the direction of the target device. However, if the target device is large to some extent, noise from the noise source is reflected by the target itself (reflected sound 31) as shown in FIG. 8 and is input to the directional acoustic sensor from the same direction as the sound emitted from the target, so noise components are It becomes the input signal including. Further, there is a wraparound sound 32 that cannot be completely erased even with directivity. Therefore, the directional acoustic sensor 1
Even if is used, a lot of noise components are included and only a signal with a poor S / N can be obtained. Therefore, a signal processing technique for extracting only a target signal from a signal including a noise component has been considered.

FIG. 9 is a diagram showing the principle of a noise cancellation technique which is one of the adaptive signal processing techniques. In the figure (a)
Indicates that the input microphone and the reference microphone are receiving sounds from the target device and the noise source. Noise from around the noise source enters the input microphone along with the sound from the target device. At this time, as shown in (b) in the figure, the noise component included in the input signal is eliminated by using the reference signal from the reference microphone that detects only noise, and only the target signal is output. It is a technology.

In practice, the noise emitted by a noise source is
Not only those that enter directly, but also those that enter after being reflected by the walls of the building etc. enter the input signal through various transmission paths.
In this way, noise enters the input signal as a noise component through a specific transfer characteristic. Therefore, this transfer function is successfully identified by using signal processing technology, and it is given as a pseudo transfer characteristic in the computer, whereby the reference signal is converted and the noise component is eliminated from the input signal to finally obtain the target. Only the signal is output. An LMS (least squares) algorithm is generally used as the transfer function identification algorithm. This is to sequentially change the transfer function from the reference signal to the input signal so as to make the output signal as small as possible. This theory is described by Bernard Widlow et al. In "Adaptive Noise Cancellation: Principles and Applications".
g: Principle and Applicati
ons) "IEEE bulletin, Vol. 63, No. 12,
1975, p. 1692-1716. In this noise canceling technique, how to take a reference signal is important. That is, the reference signal needs to have a high correlation with noise without including the sound of the target device.
This is because if the reference signal includes the sound of the target device, the effective target signal in the input signal is canceled.

In the prior art disclosed in, for example, Japanese Patent Laid-Open No. 103725/1993, noise is canceled by using a reference signal from a sensor installed near the noise source,
Only the specific sound generated from the target device is detected. FIG. 10 shows an embodiment of the above conventional noise canceling technique using a conventional directional microphone as a sensor. In the figure, an acoustic sensor 33 for obtaining the reference signal 4 is arranged near the noise source, and the noise component in the input signal 3 is removed by the computer 7 based on the reference signal 4 and output as the output signal 8. . Even with the use of the conventional directional microphone, it is possible to attenuate the reflected sound 31 and the sneak sound 32 of noise that are mixed in the input signal as a noise component.

[0006]

However, in the above-mentioned prior art, since the noise cancellation is performed by using the reference signal from the sensor installed near the noise source, there is a problem in the following cases. First, it is difficult or impossible to install the reference signal sensor in the vicinity of the noise source. For example, when the noise source moves such as a moving device or construction noise. Is. Secondly, when the noise source exists at a long distance, the cable from the reference signal sensor becomes long, and the laying work cost and the cable cost increase.

The present invention has been made in order to solve such a problem. By capturing the sound from a noise source with a directional microphone, a simple adaptive directional sound with a good S / N is obtained. An object is to provide a detection device.

[0008]

In order to solve the above problems, an adaptive directional sound detecting device according to the present invention includes a first acoustic sensor for detecting a sound generated from a target device, and a first acoustic sensor.
The second directional acoustic sensor located near the acoustic sensor and having directivity so as not to detect the sound of the target device, and the signal of the first acoustic sensor as an input signal, refer to the signal of the second directional acoustic sensor As a signal, it has adaptive noise canceling means for eliminating noise components in the input signal. further,
In the adaptive directional sound detecting device, the first acoustic sensor has directivity toward the target device. Further, in the adaptive directional sound detecting device, a plurality of second directional acoustic sensors are arranged corresponding to each noise source so as to detect only noise from the plurality of noise sources, and output from each sensor. Multi-channel adaptive noise cancellation is performed using a plurality of reference signals.

[0009]

Since the second directional acoustic sensor is located in the vicinity of the first acoustic sensor and has a directivity that does not detect the sound of the target device, the reference signal obtained by this sensor has a sound from the target device. It is possible to include noise coming from a noise source existing in a wide range other than the direction of the target device without picking up noise. Further, since the reference signal does not include the sound from the target device, reliable noise cancellation can be performed without canceling the effective target signal in the input signal in the adaptive noise canceling process in the subsequent stage. Furthermore, by providing the first acoustic sensor with directivity in the direction of the target device, it is possible to reduce noise components included in the input signal obtained by this sensor other than the direction of the target device. Further, in the adaptive directional sound detection device, a plurality of the second directional acoustic sensors are arranged corresponding to each noise source so as to detect only noise from the plurality of noise sources, and output from each sensor. By performing multi-channel adaptive noise cancellation using a plurality of reference signals, it is possible to reliably reduce noise from a plurality of noise sources.

[0010]

【Example】

Embodiment 1 FIG. FIG. 1 is a diagram showing the configuration of an adaptive directional sound detecting device according to an embodiment of the present invention. The first directional acoustic sensor 1 is arranged as an input signal microphone with directivity toward the target device. Next, without detecting the sound of the target device,
The second directional acoustic sensor 2 that detects noise generated from a wide range of noise sources is arranged on the back side of the first directional acoustic sensor as a reference signal microphone. The signals from each microphone (input signal 3 and reference signal 4) are amplified by the amplifier 5, and then the A / D converter 6
Input to and digitize. These signals are calculated by computer 7.
The noise component in the input signal is eliminated by the adaptive noise canceller processing using the LMS algorithm in the computer 7, and this is output as the output signal 8.

FIG. 2 is a diagram showing an embodiment in which the adaptive directional sound detecting device of FIG. 1 is applied to the detection of a minute sound of a rotary machine. 1m high and 1m long simulating the target device (rotary machine)
The iron plate 42 is installed vertically, and the first directional acoustic sensor 1 is installed with the directivity directed toward the center of the simulation target device. The device sound of the target device is emitted from the center of the simulated target device. The target device sound is 3 simulating a rotating machine.
A square wave of 00 Hz was used. At this time, it was verified whether the noise from the noise source could be reduced. 1 as a noise source
Random noise of ~ 6 kHz was used. The noise emitted from this noise source includes one that is reflected by the device to be simulated and enters the directional microphone 1 (reflected sound 31) and one that wraps around the directional microphone 1 (wraparound sound 32). 3 of S /
N has a high noise level of -10 dB. on the other hand,
The second directional acoustic sensor 2 for detecting the reference signal 4 is installed with directivity so as to detect only the direct sound 41 from the noise source. In this embodiment, noise is 1 to 6 kHz.
Since it covers a wide range of frequencies such as z, A / D conversion was performed at 22 kHz. A / D conversion was performed for 0.2 seconds, and after obtaining this 0.2-second continuous data, the adaptive noise canceller was executed by the software in the computer.

FIG. 3 is a diagram showing an input signal and an output signal of the adaptive directional sound detecting device. From this, by the adaptive noise cancellation, a noise reduction effect of about 20 dB was obtained in a wide range of frequency sounds from 1 to 10 kHz. As a result, the S / N of the input signal was -10 dB, but the output signal was greatly improved to 10 dB, and even a small target device sound could be detected. Figure 4 shows the LM
It is an impulse response figure at the time of adding an impulse input to the transfer function from the input identified by the S algorithm to the output. The horizontal axis represents time (msec) and the vertical axis represents weight. From this impulse response, it can be seen that both the pulse 62 corresponding to the wraparound sound 32 and the pulse 61 corresponding to the reflected sound 31 (with time delay) are successfully identified.
Therefore, it was confirmed that the noise of the noise source can be effectively reduced by the adaptive noise cancellation method.

Example 2. FIG. 5 is a diagram showing another embodiment of the sensor portion in the adaptive directional sound detecting device of FIG. In the figure, 72 and 71 are an iron pipe and a circular partition plate, which form a cylindrical body with one end closed. As the input microphone 1 representing the first acoustic sensor, a normal inexpensive omnidirectional microphone is used as it is and is fixed to the outside of the partition plate 71.
The reference microphone 2 representing the second directional acoustic sensor is a small omnidirectional microphone and is installed slightly inside from the open end of the iron pipe 72 so as to have directivity. In the case of high frequency of 5 kHz or more, sufficient directivity can be obtained with an iron pipe having a diameter of about 20 mm.

Example 3. FIG. 6 is a diagram showing another embodiment of the sensor portion in the adaptive directional sound detecting device of FIG. This is different from the above embodiment in that the partition plate 82 is provided in the middle of the iron pipe 81 instead of at one end. For the first acoustic sensor, the ordinary inexpensive omnidirectional microphone 1 is inserted deeply into the iron pipe 81, and the second directional acoustic sensor is installed near the end of the pipe with a partition plate 82 therebetween as in the above embodiment. . With such a configuration, the first acoustic sensor is characterized by having a strong directivity, which makes it possible to detect a minute sound from the target device.

Example 4. FIG. 7 is a diagram showing another embodiment of the sensor portion in the adaptive directional sound detecting device of FIG. The first acoustic sensor is installed by inserting an ordinary inexpensive omnidirectional microphone deep into the iron pipe 91. The plurality of second directional acoustic sensors (in the figure, an example of two reference microphones) is installed deep inside the iron pipes 92a and 92b, respectively. In order to obtain sufficient directivity, it is necessary to install each iron pipe in the direction of each sound source.

[0016]

The second directional acoustic sensor, which is located near the first acoustic sensor and does not detect the sound of the target device, detects the noise around the first acoustic sensor and inputs the noise of the first acoustic sensor. Since the noise component is removed from the signal, it is possible to reduce the noise coming from the noise source existing in a wide range without disturbing the sound of the target device in the input signal. This makes it possible to improve the S / N, and it may be difficult or impossible to install a reference signal sensor that moves the position of the noise source such as a moving device or construction noise near the noise source. Even in such a case, it is possible to detect with a small sensor arrangement. In particular, even when the noise source is present at a long distance and the installation work cost and the cable cost are high, the cable cost and the installation cost can be reduced. Furthermore, by providing the first acoustic sensor with directivity in the direction of the target device, noise outside the direction of the target device can be reduced and a minute sound of the target device can be collected, further improving S / N. Is possible. Further, in the adaptive directional sound detecting device, the second directional acoustic sensor is
By arranging multiple noise sources to detect only noise from multiple noise sources and performing multiple channel adaptive noise cancellation using multiple reference signals output from each sensor, It is possible to reduce the noise from the individual noise sources. As a result, a plurality of cables can be laid at the same time, laying costs can be reduced, and a compact sensor structure can be realized.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an adaptive directional sound detecting device according to an embodiment of the present invention.

FIG. 2 is a diagram showing an embodiment of FIG.

FIG. 3 is a diagram showing a noise canceling effect in one embodiment.

FIG. 4 is a diagram showing an impulse response (experimental result) which is a transfer characteristic.

FIG. 5 is a diagram showing another configuration example of the sensor unit.

FIG. 6 is a diagram showing another configuration example of the sensor unit.

FIG. 7 is a diagram illustrating another configuration example of the sensor unit.

FIG. 8 is a diagram showing an example of a conventional directional microphone.

FIG. 9 is a diagram showing the principle of a noise canceling technique.

FIG. 10 is a diagram showing an example of a conventional directional microphone + noise cancellation technique.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hiroshi Nakatani 1-2-1, Marunouchi, Chiyoda-ku, Tokyo Nihon Steel Pipe Co., Ltd.

Claims (3)

[Claims] 1. A first detecting a sound generated from a target device
An acoustic sensor, a second directional acoustic sensor installed near the first acoustic sensor and having directivity so as not to detect the sound of the target device, and a signal of the first acoustic sensor as an input signal, An adaptive directional sound detecting device, comprising: an adaptive noise canceling means for eliminating a noise component in the input signal, using a signal of the second directional acoustic sensor as a reference signal. 2. The adaptive directional sound detecting device according to claim 1, wherein the first acoustic sensor has directivity in the direction of the target device. 3. A plurality of the second directional acoustic sensors are arranged corresponding to each noise source so as to detect only noise from the plurality of noise sources, and the adaptive noise canceling means outputs from each sensor. 3. The adaptive directional sound detection device according to claim 1, wherein multi-channel adaptive noise cancellation is performed using a plurality of reference signals.